CN113094497B - Electronic health record recommendation method and shared edge computing platform - Google Patents
Electronic health record recommendation method and shared edge computing platform Download PDFInfo
- Publication number
- CN113094497B CN113094497B CN202110628271.1A CN202110628271A CN113094497B CN 113094497 B CN113094497 B CN 113094497B CN 202110628271 A CN202110628271 A CN 202110628271A CN 113094497 B CN113094497 B CN 113094497B
- Authority
- CN
- China
- Prior art keywords
- electronic health
- node
- patient
- diagnosed
- health records
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/31—Indexing; Data structures therefor; Storage structures
- G06F16/316—Indexing structures
- G06F16/322—Trees
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Abstract
The invention provides an electronic health record recommendation method and a shared edge computing platform, wherein the method comprises the following steps: obtaining context information of a patient to be diagnosed; according to the context information of the patient to be diagnosed, searching a target subspace which stores context information with similarity to the target subspace from a plurality of subspaces; according to the target subspace, finding a target node corresponding to the target subspace in the tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity; and determining a preset number of electronic health records to recommend to the patient to be diagnosed based on the plurality of electronic health records in the target node. According to the invention, similar context information is found from the subspace according to the context information of the patient to be diagnosed, then the corresponding tree node is found, and a plurality of most appropriate electronic health records are selected from the electronic health records stored in the tree node and recommended to the patient to be diagnosed, so that the method can be used for personalized diagnosis and providing support for clinical decision.
Description
Technical Field
The invention relates to the field of information recommendation, in particular to an electronic health record recommendation method and a shared edge computing platform.
Background
With the rapid development of medical technology, the kinds of diseases recognized by humans have become diversified, which makes it difficult for clinicians to diagnose.
At present, most hospitals diagnose patients to be diagnosed mainly according to the experience of medical staff, for example, the diagnosis experience of patients with the same symptoms in the past.
The diagnosis mode which basically depends on the experience of medical staff has certain subjectivity and blindness and is easy to cause misdiagnosis, and statistical data shows that the misdiagnosis can cause 10 percent of patients to die, and the misdiagnosis is the most common type of medical accidents.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides an electronic health record recommendation method based on a shared edge computing platform and the shared edge computing platform.
According to a first aspect of the present invention, there is provided an electronic health record recommendation method based on a shared edge computing platform, including: obtaining context information of a patient to be diagnosed; searching a matched target subspace from a plurality of subspaces according to the context information of the patient to be diagnosed, wherein the context information similar to the context information of the patient to be diagnosed is stored in the target subspace; according to the target subspace, finding a target node corresponding to the target subspace in a tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity; and determining a preset number of electronic health records to recommend to the patient to be diagnosed based on the plurality of electronic health records in the target node.
On the basis of the technical scheme, the invention can be improved as follows.
Optionally, one edge server is deployed in each community area, and all the edge servers and the global server together form a shared edge computing platform; and storing the context information and the electronic health records of each patient in the range of the corresponding community area on each edge server, wherein the context information of all the patients is stored in a subspace mode, and the electronic health records of all the patients are stored in a tree structure mode.
Optionally, the storing the context information of all patients in the form of subspace, and storing the electronic health records of all patients in the form of tree structure includes: performing similarity clustering on the context information of all patients, and storing the context information with similarity into the same subspace; carrying out similarity clustering on the electronic health records of all patients, and storing the electronic health records with similarity in the same node of a binary tree structure; each subspace and each node have a one-to-one correspondence relationship.
Optionally, the determining, based on the plurality of electronic health records in the target node, that a preset number of electronic health records are recommended to the patient to be diagnosed includes: for any one electronic health record in the target node, obtaining a plurality of recommended feedback results of the any one electronic health record; determining a weight for the any one electronic health record based on a plurality of recommended feedback results for the any one electronic health record; and sequencing all the electronic health records according to the weight of each electronic health record, and recommending the preset number of the electronic health records which are sequenced at the top to the patient to be diagnosed.
Optionally, the preset number can be adjusted according to the current time period of the patient to be diagnosed.
Optionally, before determining that a preset number of electronic health records are recommended to a patient to be diagnosed based on a plurality of electronic health records in the target node, the method includes: determining, based on the target node, a unique path from a root node to the target node in a tree structure; securing the reward value of the electronic health record stored in each node on the unique path.
Optionally, the securing the reward value of the electronic health record stored in each node on the unique path includes: determining the reward value of any one electronic health record according to the recommendation feedback result of any one electronic health record stored in each node; adding the reward values of all the electronic health records stored in any one node to obtain the total reward value of any one node; and adding Laplace noise into the total reward value, and calculating a corresponding expected reward value.
Optionally, the method further includes: taking the context information of all patients and the electronic health records of all patients stored in each edge server as a local storage model of the edge server; after the patient to be diagnosed is diagnosed, updating the local storage model of the corresponding edge server to obtain a corresponding intermediate storage module, and sending the intermediate storage module to the global server; the global server updates a local global storage model according to the intermediate storage model sent by each edge server and sends the updated global storage model to each edge server; and the edge server updates the local intermediate storage model based on the updated global storage model to obtain a final updated local storage model so as to search the corresponding subspace and the node in the updated local storage model.
According to a second aspect of the present invention, there is provided a shared edge computing platform comprising a plurality of edge servers and a global server, each edge server being connected to the global server via a communications network; wherein, an edge server is deployed in each community area range, all edge servers and a global server jointly form a shared edge computing platform, one of the edge servers is a management node, and the management node is used for: obtaining context information of a patient to be diagnosed; searching a matched target subspace from a plurality of subspaces according to the context information of the patient to be diagnosed, wherein the context information similar to the context information of the patient to be diagnosed is stored in the target subspace; according to the target subspace, finding a target node corresponding to the target subspace in a tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity; and determining a preset number of electronic health records to recommend to the patient to be diagnosed based on the plurality of electronic health records in the target node.
Optionally, the determining, based on the plurality of electronic health records in the target node, that a preset number of electronic health records are recommended to the patient to be diagnosed includes: for any one electronic health record in the target node, obtaining a plurality of recommended feedback results of the any one electronic health record; determining a weight for the any one electronic health record based on a plurality of recommended feedback results for the any one electronic health record; and sequencing all the electronic health records according to the weight of each electronic health record, and recommending the preset number of the electronic health records which are sequenced at the top to the patient to be diagnosed.
According to the electronic health record recommendation method based on the shared edge computing platform and the shared edge computing platform, similar context information is found from the subspace according to the context information of the patient to be diagnosed, then the corresponding tree node is found, and the most appropriate electronic health records are selected from the electronic health records stored in the tree node and recommended to the patient to be diagnosed, so that the method can be used for personalized diagnosis and providing support for clinical decision.
Drawings
FIG. 1 is a flowchart of an electronic health record recommendation method based on a shared edge computing platform according to the present invention;
FIG. 2 is a schematic diagram of a shared edge computing platform.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
Fig. 1 is a flowchart of an electronic health record recommendation method based on a shared edge computing platform, as shown in fig. 1, the method includes: 101. obtaining context information of a patient to be diagnosed; 102. searching a matched target subspace from a plurality of subspaces according to the context information of the patient to be diagnosed, wherein the context information similar to the context information of the patient to be diagnosed is stored in the target subspace; 103. according to the target subspace, finding a target node corresponding to the target subspace in a tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity; 104. and determining a preset number of electronic health records to recommend to the patient to be diagnosed based on the plurality of electronic health records in the target node.
It can be understood that, at present, the diagnosis of a patient to be diagnosed is generally judged according to the experience of medical staff, and the judgment mode is subjective and has low accuracy. Based on this, in order to adapt to the rapid increase of clinical diagnosis demand, an Electronic Health Record (EHR) is utilized as an auxiliary method to help doctors make accurate and personalized clinical decisions.
Different patients have different situations (e.g., physical conditions) that require personalized diagnosis, and the diagnosis error rate can be greatly reduced by the patient's environment. In a medical diagnostic system, contextual information for a patient will be collected and uploaded to a server that recommends an EHR for the patient to assist in diagnosis.
The embodiment of the invention provides a shared edge computing platform, and a cooperative mobile edge computing platform MEC is a new computing paradigm, requires data processing on an Edge Server (ES) of a network, and can communicate with the ES within a shorter propagation distance compared with the traditional cloud computing, so that relatively lower delay is realized. In addition, information will be shared between the ES to improve the recommendation utility of the system. Furthermore, such paradigm may distribute data processing work over different data centers and devices, which makes it difficult for hackers to attack the entire network, thereby enhancing security of privacy protection.
The method comprises the steps of storing context information of each historical patient in each subspace, storing the context information with similarity in the same subspace, storing Electronic Health Records (EHRs) of each historical patient in a tree structure, and storing the electronic health records with similarity in the same node of the tree structure. Wherein each subspace has a corresponding relationship with each node in the tree structure.
When a patient to be diagnosed needs to be diagnosed, the context information of the patient to be diagnosed is obtained first, and based on the similarity principle, a target subspace matched with the context information of the patient to be diagnosed is found from all subspaces, that is, the context information stored in the target subspace has similarity with the context information of the patient to be diagnosed.
And finding a target subspace, finding a target node corresponding to the target subspace according to the corresponding relation between the subspace and the tree structure node, wherein a plurality of electronic health records are stored in the target node, and selecting a proper preset number of electronic health records from all the electronic health records of the target node to recommend to the patient to be diagnosed.
According to the invention, similar context information is found from the subspace according to the context information of the patient to be diagnosed, then the corresponding tree node is found, and a plurality of most appropriate electronic health records are selected from the electronic health records stored in the tree node and recommended to the patient to be diagnosed, so that the method can be used for personalized diagnosis and providing support for clinical decision.
In a possible embodiment mode, one edge server is deployed in each community area range, and all the edge servers and the global server jointly form a shared edge computing platform; and storing the context information and the electronic health records of each patient in the range of the corresponding community area on each edge server, wherein the context information of all the patients is stored in a subspace mode, and the electronic health records of all the patients are stored in a tree structure mode.
It is appreciated that embodiments of the present invention provide a shared edge computing platform that includes a plurality of Edge Servers (ESs) and a global server. Considering that the storage space of the edge server ES is limited, it is not practical to put all EHRs stored in the cloud into one ES for recommendation. Therefore, it is necessary to decide which EHRs to place in each ES, i.e., determine the best service placement scheme for each ES. Embodiments of the present invention contemplate an edge computing network architecture, i.e. a shared edge computing voucher, in which an edge server ES is deployed in edge computing in the vicinity of a hospital. Each ES covers a different community, from which context information of the patient is collected and processed to recommend the relevant electronic health record EHR to the patient as a reference.
It should be noted that within the framework of the edge computing network architecture, these edge servers ES are themselves trustworthy and do not expose the patient's situation to others. It is assumed that the network consists of M ES (collaborative mobile edge computing) around the hospital, which can be used to diagnose patients in the community. Considering the edge computing network structure as a discrete time system, and dividing the total timeline into T time segments, with T = {1, 2NIs the index. At the beginning of each time period, assuming that there is a patient needing diagnosis, whose corresponding context information is pt, in order for an EHR in the ES to diagnose the patient, N = {1, 2.
The context space model of the patient is described below, each patient having a disease to be diagnosed, and the context of the patient is used as the basic input for the diagnosis, denoted by p. The context information p is formulated as a kappa-p dimensional vector, each entry of which is characteristic of patient health related information (e.g., lifestyle, disease history, etc.). Without loss of generality, assume that each entry is normalized to [0, 1 ]. For example, when κ P =3 and a situation P = [0.9, 0.8, 1] for a patient arriving at space P, where 0.9 indicates that the patient's lifestyle is reasonably good, 0.8 indicates that the patient is coughing severely, and 1 indicates that the patient has a particular disease. In this way, the contextual information of the patient can be quantified and analyzed by the server for diagnosis.
Without loss of generality, a kd-dimensional vector may be used to represent any EHR in the ES, each entry of which is a function in the EHR. Because of the large number of EHRs in the ES, a binary EHR tree structure group is created that groups those EHRs-like in the nodes in the tree. Thus, the problem of big data analysis can be solved if the dimension kd is really high, for a particular patient with a particular context, one tends to choose a node that fits the recommended EHR to help the patient make a diagnosis. For example, certain EHRs are matched to the condition of the patient and, therefore, should be recommended to the patient to aid diagnosis, since the patient is most likely to record the type of disease in these EHRs. The kth node with height l is defined as (l, k), so intuitively speaking, the left child node and the right child node are (l +1, 2 k) and (l +1, 2k + 1), respectively. We denote lY as the depth of the tree. When certain conditions are met, the tree structure will be expanded by dividing the nodes into two groups of approximately the same size. Based on these vectors of EHRs, when a node needs to be partitioned, they can be divided into two groups. Notably, the EHR tree structure can be extended indefinitely, which is an ideal way to handle large data scenarios.
To summarize, when the context information and the electronic health record of each patient are stored in each edge server ES, the context information of each patient is stored in a subspace mode, specifically, the context information of all patients is subjected to similarity clustering, and the context information with similarity is stored in the same subspace. The electronic health records of each patient are stored in a tree structure, specifically, the electronic health records of all patients are subjected to similarity clustering, and the electronic health records with similarity are stored in the same node of the tree structure. The tree structure comprises a root node and a leaf node, wherein a one-to-one correspondence relationship exists between a subspace for storing the context information of the patient and the nodes for storing the electronic health records, namely, the one-to-one correspondence relationship is provided for the one-class context information and the one-class electronic health records corresponding to the one-class context information.
In a possible embodiment, determining that a preset number of electronic health records are recommended to a patient to be diagnosed based on a plurality of electronic health records in a target node includes: for any electronic health record in a target node, obtaining a plurality of recommended feedback results of the any electronic health record; determining a weight for the any one electronic health record based on a plurality of recommended feedback results for the any one electronic health record; and sequencing all the electronic health records according to the weight of each electronic health record, and recommending the preset number of the electronic health records which are sequenced at the top to the patient to be diagnosed.
It can be understood that a target subspace with similar contexts is found according to the context information of the patient to be diagnosed, a corresponding target node is found according to the subspace, and for all the electronic health records stored in the target node, a plurality of appropriate electronic health records are selected from the electronic health records and recommended to the patient to be diagnosed.
Specifically, for any electronic health record, after the electronic health record is recommended to the patient, a feedback result of the patient is received, and the feedback result shows the accuracy of the recommendation and the diagnosis effect. It should be noted that, if one electronic health record is recommended to a plurality of patients, recommendation feedback results of the plurality of patients are received. For any electronic health record stored in a tree structure, obtaining a plurality of recommendation feedback results corresponding to the electronic health record, and giving a weight to the electronic health record according to the recommendation feedback results, wherein the more accurate the recommendation is, the greater the diagnosis utility after recommendation is, the greater the weight is; conversely, the less weight is assigned to the electronic health record.
And sorting all the electronic health records stored in the target node from large to small according to the weight, and recommending the preset number of electronic health records which are sorted in the front to the patient to be diagnosed.
The preset number can be adjusted according to the current time period of the patient to be diagnosed, that is, the maximum number of recommended electronic health records for the patient to be diagnosed can be set for different time periods. When the context information of the patient to be diagnosed is obtained, the current time is also obtained, and the maximum number of the recommended electronic health records for the patient to be diagnosed is determined based on the time period of the current time, namely the number of the recommended electronic health records for the patient to be diagnosed cannot exceed the set maximum number.
In a possible embodiment, before determining that a preset number of electronic health records are recommended to a patient to be diagnosed based on a plurality of electronic health records in the target node, the method includes: determining, based on a target node, a unique path from a root node to the target node in a tree structure; securing the reward value of the electronic health record stored in each node on the unique path.
It will be appreciated that there are important ways to protect the privacy of patient medical data, and that privacy protection is currently available in a variety of ways, and it is quite common and useful to apply anonymity. However, this comes at the expense of the potential utility of patient data in a high dimension. Cryptography is also a very popular method of ensuring security, but it results in high communication costs and computational complexity, which can greatly affect large data analysis. Therefore, as a preferred method, the embodiment of the present invention deploys differential privacy in the medical diagnosis system, and performs privacy protection on the electronic health record recommended to the patient to be diagnosed.
Specifically, in the above embodiment, a target node in the tree structure is found, and based on the target node, a unique path from a root node of the tree structure to the target node is determined, where the unique path includes a plurality of nodes, and each node stores a plurality of electronic health records.
The reward values of all electronic health records stored in each node on the unique path are secured, mainly for privacy protection.
In one possible embodiment, securing the reward value of the electronic health record stored in each node on the unique path includes: determining the reward value of any electronic health record according to the recommendation feedback result of any electronic health record stored in each node; adding the reward values of all the electronic health records stored in any one node to obtain the total reward value of any one node; laplace noise is added to the total reward value, and a corresponding expected reward value is calculated.
It is to be understood that for any one of the electronic health records stored in each node, the reward value for any one of the electronic health records is determined based on the recommendation feedback results for that one of the electronic health records. For all electronic health records in any one node, the reward values for each electronic health record are summed to obtain a total reward value for that any one node. And privacy protection is performed on the total reward value of each node, specifically, laplace noise is added into the total reward value, a corresponding expected reward value is calculated, and privacy protection is performed on the electronic health record recommended to the patient to be diagnosed.
In a possible implementation manner, the method further includes: taking the context information of all patients and the electronic health records of all patients stored in each edge server as a local storage model of the edge server; after the patient to be diagnosed is diagnosed, updating the local storage model of the corresponding edge server to obtain a corresponding intermediate storage module, and sending the intermediate storage module to the global server; the global server updates a local global storage model according to the intermediate storage model sent by each edge server and sends the updated global storage model to each edge server; and the edge server updates the local intermediate storage model based on the updated global storage model to obtain a final updated local storage model so as to search the corresponding subspace and the node in the updated local storage model.
It can be understood that, for the shared edge computing platform provided by the embodiment of the present invention, as described above, each edge server stores the context information and the electronic health record of all patients in the local community area, and the information stored in each edge server is referred to as a local storage model. After the diagnosis of the patient to be diagnosed is completed, the local storage model is updated, that is, the data stored in the local storage model is updated, and the updated local storage model is called an intermediate storage model. And each edge server sends the updated intermediate storage model to the global server, wherein the global server stores the data of all patients in all community areas, including context information and electronic health records, namely, the data stored by all edge servers is summarized by the data in the global storage model in the global server.
And when the global server receives the intermediate storage model sent by each edge server, updating the local global storage model based on all the intermediate storage models, and sending the updated global storage model to each edge server. And each edge server updates the local intermediate storage model again according to the updated global storage model to obtain a final updated local storage model. In particular, the various subspaces that require the edge server to locally store context information and the various nodes of the tree structure that store electronic health records are also in need of updating. And when waiting for the next patient to be diagnosed, searching the corresponding subspace and the node according to the updated subspace node.
Fig. 2 is a schematic structural diagram of a shared edge computing platform according to an embodiment of the present invention, as shown in fig. 2, the shared edge computing platform includes a plurality of edge servers 20 and a global server 21, each edge server 20 is connected to the global server 21 through a communication network; wherein, one edge server 20 is deployed in each community area, and all the edge servers 20 and the global server 21 together form a shared edge computing platform. One of the edge servers 20 is a management node, and the management node is configured to: obtaining context information of a patient to be diagnosed; searching a matched target subspace from the plurality of subspaces according to the context information of the patient to be diagnosed, wherein the context information similar to the context information of the patient to be diagnosed is stored in the target subspace; according to the target subspace, finding a target node corresponding to the target subspace in the tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity; and determining a preset number of electronic health records to recommend to the patient to be diagnosed based on the plurality of electronic health records in the target node.
Wherein, based on a plurality of electronic health records in the target node, determining that a preset number of electronic health records are recommended to a patient to be diagnosed, comprising: for any electronic health record in the target node, obtaining a plurality of recommended feedback results of any electronic health record; determining a weight for any of the electronic health records based on the plurality of recommended feedback results for any of the electronic health records; and sequencing all the electronic health records according to the weight of each electronic health record, and recommending the preset number of the electronic health records which are sequenced at the top to the patient to be diagnosed.
It can be understood that, as for the technical features related to how to perform the electronic health record recommendation on the shared edge computing platform provided in the embodiment of the present invention, reference may be made to the technical features related to the electronic health record recommendation method based on the shared edge computing platform provided in the foregoing embodiments, and details are not described here again.
According to the electronic health record recommendation method based on the shared edge computing platform and the shared edge computing platform, similar context information is found from the subspace according to the context information of the patient to be diagnosed, then the corresponding tree node is found, and the most appropriate electronic health records are selected from the electronic health records stored in the tree node and recommended to the patient to be diagnosed, so that the method can be used for personalized diagnosis and providing support for clinical decision.
In addition, when a plurality of proper electronic health records are selected from all the electronic health records of the target node and recommended to the patient to be diagnosed, the recommendation feedback result of each electronic health record is considered, the weight of the electronic health record is determined according to the recommendation feedback result, the proper electronic health record is recommended to the patient to be diagnosed according to the weight of each electronic health record, and the accuracy of recommending the electronic health records is improved.
Moreover, the electronic health records recommended to the patient to be diagnosed are subjected to privacy protection, and the safety of the whole shared edge computing platform is enhanced.
Finally, for all edge servers in the shared edge computing platform, the edge servers are not communicated with each other directly, data exchange is not performed directly, and data exchange between the edge servers is realized through the global server, so that the problem of safety caused by data exchange between the edge servers is avoided, and the safety of the shared edge computing platform is improved.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (8)
1. An electronic health record recommendation method based on a shared edge computing platform is characterized by comprising the following steps:
obtaining context information of a patient to be diagnosed;
searching a matched target subspace from a plurality of subspaces according to the context information of the patient to be diagnosed, wherein the context information similar to the context information of the patient to be diagnosed is stored in the target subspace;
according to the target subspace, finding a target node corresponding to the target subspace in a tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity;
determining a preset number of electronic health records to recommend to a patient to be diagnosed based on the plurality of electronic health records in the target node;
wherein before determining that a preset number of electronic health records are recommended to a patient to be diagnosed based on a plurality of electronic health records in the target node, the method comprises:
determining, based on the target node, a unique path from a root node to the target node in a tree structure;
securing a reward value of an electronic health record stored in each node on the unique path;
the securing the reward value of the electronic health record stored in each node on the unique path includes:
determining the reward value of any one electronic health record according to the recommendation feedback result of any one electronic health record stored in each node;
adding the reward values of all the electronic health records stored in any one node to obtain the total reward value of any one node;
and adding Laplace noise into the total reward value, and calculating a corresponding expected reward value.
2. The electronic health record recommendation method of claim 1,
deploying an edge server in each community area range, wherein all the edge servers and the global server jointly form a shared edge computing platform;
and storing the context information and the electronic health records of each patient in the range of the corresponding community area on each edge server, wherein the context information of all the patients is stored in a subspace mode, and the electronic health records of all the patients are stored in a tree structure mode.
3. The electronic health record recommendation method according to claim 2, wherein storing the context information of all patients in a subspace and storing the electronic health records of all patients in a tree structure comprises:
performing similarity clustering on the context information of all patients, and storing the context information with similarity into the same subspace;
carrying out similarity clustering on the electronic health records of all patients, and storing the electronic health records with similarity in the same node of a binary tree structure;
each subspace and each node have a one-to-one correspondence relationship.
4. The electronic health record recommendation method according to any one of claims 1-3, wherein determining a preset number of electronic health records to recommend to a patient to be diagnosed based on a plurality of electronic health records in the target node comprises:
for any one electronic health record in the target node, obtaining a plurality of recommended feedback results of the any one electronic health record;
determining a weight for the any one electronic health record based on a plurality of recommended feedback results for the any one electronic health record;
and sequencing all the electronic health records according to the weight of each electronic health record, and recommending the preset number of the electronic health records which are sequenced at the top to the patient to be diagnosed.
5. The electronic health record recommendation method of claim 4, wherein the preset number is adjustable according to a time period currently in which a patient to be diagnosed is located.
6. The electronic health record recommendation method of claim 1, further comprising:
taking the context information of all patients and the electronic health records of all patients stored in each edge server as a local storage model of the edge server;
after the patient to be diagnosed is diagnosed, updating the local storage model of the corresponding edge server to obtain a corresponding intermediate storage module, and sending the intermediate storage module to the global server;
the global server updates a local global storage model according to the intermediate storage model sent by each edge server and sends the updated global storage model to each edge server;
and the edge server updates the local intermediate storage model based on the updated global storage model to obtain a final updated local storage model so as to search the corresponding subspace and the node in the updated local storage model.
7. A shared edge computing platform comprising a plurality of edge servers and a global server, each edge server connected to the global server via a communications network; wherein, an edge server is deployed in each community area range, all edge servers and a global server jointly form a shared edge computing platform, one of the edge servers is a management node, and the management node is used for:
obtaining context information of a patient to be diagnosed;
searching a matched target subspace from a plurality of subspaces according to the context information of the patient to be diagnosed, wherein the context information similar to the context information of the patient to be diagnosed is stored in the target subspace;
according to the target subspace, finding a target node corresponding to the target subspace in a tree structure, wherein each node of the tree structure stores a plurality of electronic health records with similarity;
determining a preset number of electronic health records to recommend to a patient to be diagnosed based on the plurality of electronic health records in the target node;
wherein before determining that a preset number of electronic health records are recommended to a patient to be diagnosed based on a plurality of electronic health records in the target node, the method comprises:
determining, based on the target node, a unique path from a root node to the target node in a tree structure;
securing a reward value of an electronic health record stored in each node on the unique path;
the securing the reward value of the electronic health record stored in each node on the unique path includes:
determining the reward value of any one electronic health record according to the recommendation feedback result of any one electronic health record stored in each node;
adding the reward values of all the electronic health records stored in any one node to obtain the total reward value of any one node;
and adding Laplace noise into the total reward value, and calculating a corresponding expected reward value.
8. The shared edge computing platform of claim 7, wherein determining that a preset number of electronic health records are recommended to a patient to be diagnosed based on a plurality of electronic health records in the target node comprises:
for any one electronic health record in the target node, obtaining a plurality of recommended feedback results of the any one electronic health record;
determining a weight for the any one electronic health record based on a plurality of recommended feedback results for the any one electronic health record;
and sequencing all the electronic health records according to the weight of each electronic health record, and recommending the preset number of the electronic health records which are sequenced at the top to the patient to be diagnosed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110628271.1A CN113094497B (en) | 2021-06-07 | 2021-06-07 | Electronic health record recommendation method and shared edge computing platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110628271.1A CN113094497B (en) | 2021-06-07 | 2021-06-07 | Electronic health record recommendation method and shared edge computing platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113094497A CN113094497A (en) | 2021-07-09 |
| CN113094497B true CN113094497B (en) | 2021-09-14 |
Family
ID=76664558
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110628271.1A Active CN113094497B (en) | 2021-06-07 | 2021-06-07 | Electronic health record recommendation method and shared edge computing platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113094497B (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102985924A (en) * | 2011-02-14 | 2013-03-20 | 松下电器产业株式会社 | Similar case retrieval device and similar case retrieval method |
| US9905112B1 (en) * | 2015-05-18 | 2018-02-27 | HCA Holdings, Inc. | Contextual assessment of current conditions |
| CN108877880A (en) * | 2018-06-29 | 2018-11-23 | 清华大学 | Patient's similarity measurement device and method based on case history text |
| CN109215754A (en) * | 2018-09-10 | 2019-01-15 | 平安科技（深圳）有限公司 | Medical record data processing method, device, computer equipment and storage medium |
| CN109599182A (en) * | 2019-02-18 | 2019-04-09 | 湖南金特尔信息技术有限公司 | A kind of gene target therapeutic scheme recommender system and related system |
| CN110249392A (en) * | 2018-08-20 | 2019-09-17 | 深圳市全息医疗科技有限公司 | Intelligent assisting in diagnosis and treatment system and method |
| CN111105880A (en) * | 2019-11-27 | 2020-05-05 | 陕西医链区块链集团有限公司 | Hierarchical search system and method for case calling |
| CN111326224A (en) * | 2013-12-04 | 2020-06-23 | M·奥利尼克 | Medical treatment computer planning method and system with mass medical analysis |
| CN111464650A (en) * | 2020-04-03 | 2020-07-28 | 北京绪水互联科技有限公司 | Data analysis method, equipment, system and storage medium |
| CN112289437A (en) * | 2020-10-28 | 2021-01-29 | 南通市第一人民医院 | Diabetes adjuvant therapy cloud platform system based on edge computing framework |
| CN112800248A (en) * | 2021-01-19 | 2021-05-14 | 天河超级计算淮海分中心 | Similar case retrieval method, similar case retrieval device, computer equipment and storage medium |
| CN112837770A (en) * | 2021-01-21 | 2021-05-25 | 北京理工大学 | Privacy protection similar medical record query method in large-scale electronic medical system |
| CN112908493A (en) * | 2021-01-28 | 2021-06-04 | 壹健康健康产业(深圳)有限公司 | Online inquiry processing method, device, medium and terminal equipment |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120232930A1 (en) * | 2011-03-12 | 2012-09-13 | Definiens Ag | Clinical Decision Support System |
| US10123748B2 (en) * | 2014-10-30 | 2018-11-13 | International Business Machines Corporation | Active patient risk prediction |
| CN105608389B (en) * | 2015-10-22 | 2018-04-20 | 广西师范大学 | The difference method for secret protection of medical data issue |
| CN107526975A (en) * | 2017-08-10 | 2017-12-29 | 中国人民大学 | A kind of method based on difference secret protection decision tree |
| CN109241774B (en) * | 2018-09-19 | 2020-11-10 | 华中科技大学 | Differential privacy space decomposition method and system |
| US20200243180A1 (en) * | 2019-01-25 | 2020-07-30 | International Business Machines Corporation | Optimized medication recommendation using a neural network |
| CN110134879B (en) * | 2019-03-06 | 2021-06-25 | 辽宁工业大学 | Interest point recommendation algorithm based on differential privacy protection |
| CN111125764B (en) * | 2019-12-23 | 2022-02-15 | 华中科技大学 | Privacy protection-oriented user track generation method and system |
| CN112652375A (en) * | 2021-01-08 | 2021-04-13 | 北京工业大学 | Medicine recommendation method and device, electronic equipment and storage medium |
-
2021
- 2021-06-07 CN CN202110628271.1A patent/CN113094497B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102985924A (en) * | 2011-02-14 | 2013-03-20 | 松下电器产业株式会社 | Similar case retrieval device and similar case retrieval method |
| CN111326224A (en) * | 2013-12-04 | 2020-06-23 | M·奥利尼克 | Medical treatment computer planning method and system with mass medical analysis |
| US9905112B1 (en) * | 2015-05-18 | 2018-02-27 | HCA Holdings, Inc. | Contextual assessment of current conditions |
| CN108877880A (en) * | 2018-06-29 | 2018-11-23 | 清华大学 | Patient's similarity measurement device and method based on case history text |
| CN110249392A (en) * | 2018-08-20 | 2019-09-17 | 深圳市全息医疗科技有限公司 | Intelligent assisting in diagnosis and treatment system and method |
| CN109215754A (en) * | 2018-09-10 | 2019-01-15 | 平安科技（深圳）有限公司 | Medical record data processing method, device, computer equipment and storage medium |
| CN109599182A (en) * | 2019-02-18 | 2019-04-09 | 湖南金特尔信息技术有限公司 | A kind of gene target therapeutic scheme recommender system and related system |
| CN111105880A (en) * | 2019-11-27 | 2020-05-05 | 陕西医链区块链集团有限公司 | Hierarchical search system and method for case calling |
| CN111464650A (en) * | 2020-04-03 | 2020-07-28 | 北京绪水互联科技有限公司 | Data analysis method, equipment, system and storage medium |
| CN112289437A (en) * | 2020-10-28 | 2021-01-29 | 南通市第一人民医院 | Diabetes adjuvant therapy cloud platform system based on edge computing framework |
| CN112800248A (en) * | 2021-01-19 | 2021-05-14 | 天河超级计算淮海分中心 | Similar case retrieval method, similar case retrieval device, computer equipment and storage medium |
| CN112837770A (en) * | 2021-01-21 | 2021-05-25 | 北京理工大学 | Privacy protection similar medical record query method in large-scale electronic medical system |
| CN112908493A (en) * | 2021-01-28 | 2021-06-04 | 壹健康健康产业(深圳)有限公司 | Online inquiry processing method, device, medium and terminal equipment |
Non-Patent Citations (1)
| Title |
|---|
| "Improving Health Care Outcomes through Personalized Comparisons of Treatment Effectiveness Based on Electronic Health Records";Sharona Hoffman;《The Journal of Law, Medicine & Ethics》;20111231;第424-436页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113094497A (en) | 2021-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110462604A (en) | The data processing system and method for association internet device are used based on equipment | |
| CN113535974B (en) | Diagnostic recommendation method and related device, electronic equipment and storage medium | |
| US10671646B2 (en) | Methods and systems for linking data records from disparate databases | |
| KR20130062442A (en) | Method and system for recommendation using style of collaborative filtering | |
| US20140006166A1 (en) | System and method for determining offers based on predictions of user interest | |
| US20160125159A1 (en) | System for management of health resources | |
| CN109788489A (en) | A kind of base station planning method and device | |
| Folino et al. | A comorbidity-based recommendation engine for disease prediction | |
| CN108242266A (en) | Auxiliary diagnostic equipment and method | |
| CN110322356A (en) | The medical insurance method for detecting abnormality and system of dynamic multi-mode are excavated based on HIN | |
| CN107967487A (en) | A kind of colliding data fusion method based on evidence distance and uncertainty | |
| CN113315752B (en) | Intelligent medical attack tracing method based on block chain and medical big data system | |
| CN112132624A (en) | Medical claims data prediction system | |
| CN114065856B (en) | Doctor recommendation method, device and equipment based on doctor portrait and storage medium | |
| CN114639483A (en) | Electronic medical record retrieval method and device based on graph neural network | |
| CN113192586A (en) | Intelligent medical file sharing method based on block chain and medical big data system | |
| CN113539460A (en) | Intelligent diagnosis guiding method and device for remote medical platform | |
| CN113853599A (en) | Efficient freshness crawling scheduling | |
| CN117078048A (en) | Digital twinning-based intelligent city resource management method and system | |
| CN111259167B (en) | User request risk identification method and device | |
| CN114942947A (en) | Follow-up visit data processing method and system based on intelligent medical treatment | |
| CN114091610A (en) | Intelligent decision method and device | |
| US20180285758A1 (en) | Methods for creating and analyzing dynamic trail networks | |
| CN117370565A (en) | Information retrieval method and system | |
| CN116569194A (en) | Joint learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |